Moonshot AI’s Kimi K3, a 2.8-trillion-parameter open-weight model with 1M-token context, is now on Amazon Bedrock

Moonshot AI’s Kimi K3 is now available on Amazon Bedrock, providing an additional open-weight option for coding and knowledge work. The model is described by Moonshot AI as its most capable to date and the first open model to reach 2.8 trillion parameters.

Kimi K3 includes native vision capabilities, a 1-million-token context window, and explicit prompt caching, which is intended to reduce latency and input costs when reusing context across multiple model calls.

According to Moonshot AI, Kimi K3 achieves an approximate 2.5x improvement in scaling efficiency over its predecessor, Kimi K2, positioning it for long-running workflows that require sustained context across large code repositories, documents, and images.

The launch reflects AWS’s sustained investment in open-weight models on Amazon Bedrock, which since 2025 has added dozens of open-weight models from providers including DeepSeek, Google, MiniMax, Mistral AI, Moonshot AI, NVIDIA, OpenAI, and Qwen.